AWS for SAP

Amazon EC2 High Memory Instances now available for on-demand usage

Introduction
SAP customers continue to leverage AWS as their platform of choice and innovation. Some are in the early stages of their SAP cloud journeys and are focused on executing their migration. Others have hardened their SAP systems on AWS and are innovating around their core business processes with advanced AWS services. For example, Zalando has reinvented their data and analytics architecture to improve customer experience, Invista is using AI/ML with ECC to drive better manufacturing outcomes, and Volkswagen has integrated their S/4HANA systems with AWS IoT as part of their Digital Production Platform initiative. But even for SAP customers who are transforming with these innovative solutions at the line of business and technical use case level, the underlying infrastructure supporting their SAP systems remains a critical consideration.

That’s why today, in response to customer demand, we’re excited to announce general availability of new instance sizes within our Amazon EC2 High Memory Instances family with 6, 9, and 12TB of memory as well as support for on-demand consumption. This allows you to take advantage of hourly pricing and EC2 Savings Plans, while also giving you more options for supporting use cases with temporary infrastructure needs.

Before diving into the details of this launch, I’d like to briefly cover how we got here, why we continue to invest in infrastructure for SAP, and how AWS customers are taking advantage of these investments.

Why SAP customers count on the AWS Global Infrastructure

We have been supporting SAP customers since 2011. Over this 10+ year journey, I have watched customer adoption patterns move from early dev/test workloads to becoming the net new normal for production SAP deployments and large data center exit migrations. One of the key reasons we’re seeing this is because customers realize they need to protect themselves from disruption to mission-critical business processes, and the AWS Global Infrastructure makes it a lot easier to architect for availability and resilience than has been possible on-premises.

Take, for example, Bristol Myers Squibb. They actually built high availability and disaster recovery testing into their ECC to S/4HANA on AWS migration process. Now, they take advantage of the AWS multi-AZ design for high availability and a multi-region design concept for disaster recovery, which helps them meet their needs for uptime and system resilience. Or, for example the US Navy, who last year migrated their SAP ERP system—which supports 72,000 users and the movement of some $70B worth of goods and services—to AWS to improve availability, and scalability while helping them make more timely and informed decisions. You could also look at Zalando. While they leverage 36+ AWS services in concert with S/4HANA on AWS to support their continued growth and transformation, they still recognize the role that core infrastructure plays in their success. “AWS infrastructure and broad set of services continue to be key enablers of not just our SAP strategy, but to our broader strategy to deliver better experiences for our customers, operate more efficiently, and grow the business” said Yuriy Volesenko, former Director of Enterprise Applications & Architectures at Zalando SE. Customer feedback drives nearly everything we do at AWS, and it’s been feedback like this that has validated our continued focus on delivering the world’s best infrastructure purpose-built for SAP workloads.

The creation of High Memory Instances is one such example. After we released our Amazon EC2 X1 instance in 2016—the first cloud instance purpose-built for large in-memory databases like SAP HANA—customers pushed us to go even bigger to help them support their multi-terabyte HANA-based systems in production on AWS. In 2018, we released High Memory Instances in response, which now offer up to 24TB of memory in a single instance. As High Memory Instances have grown in popularity, customers have asked for additional flexibility including on-demand usage and savings plans.

Introducing new High Memory instances with Nitro Hypervisor

Today’s announcement of new 6, 9, and 12 TB High Memory instance sizes delivers on that request, allowing you to realize the same security, performance, and flexibility benefits of the AWS Nitro System for large SAP HANA workloads, while supporting new use cases with hourly pricing. For example, customers may want to build temporary systems for testing new features, fixing bugs, or building sandbox environments. These new instance sizes makes it easier to support some of these uses cases with on-demand infrastructure requirements. In addition, when customer wants to continue their usage for extended period, support for EC2 Savings Plans allows customers to lower their compute costs by up to 72% compared to on-demand pricing.

Additionally some customers have told us they don’t need all the CPU capacity offered on the 6TB instances. To help, we are also introducing a new non-hyperthreaded version of our 6TB instance with fewer vCPUs. This instance offers 6TB of memory with 224 vCPUs still capable of strong compute performance, but at a lower price.

A brief overview of all new sizes is as follows:

Instance size

Memory (GiB)

vCPU

Network Bandwidth (Gbps)

Dedicated EBS Bandwidth (Gbps)

SAPS

u-6tb1.56xlarge 6,144 224* 100 38 380,770
u-6tb1.112xlarge 6,144 448 100 38 475,500
u-9tb1.112xlarge 9,216 448 100 38 475,500
u-12tb1.112xlarge 12,288 448 100 38 475,500

* non-hyperthreaded cores

Like all of our instances, these new High Memory Instances will provide a consistent, cloud-native experience that is seamlessly integrated with other AWS services like Amazon EBS. They are also based on the AWS Nitro System, which I referenced before. If you are unfamiliar with Nitro, it’s the underlying platform for our latest generation of EC2 instances, which provides a number of benefits:

  • Performance: The Nitro System delivers all the compute and memory resources of the host hardware to your Amazon EC2 instances. As a result, it eliminates the performance hit that typically comes from a hypervisor layer. Additionally, our dedicated Nitro Cards offer high-speed networking, block storage, and I/O acceleration. Because of these Nitro benefits, these new instance sizes offer performance with ~1% variance when compared to our existing bare metal High Memory instances. As part of this launch, we have published several SAP benchmarks showcasing the performance of these new instance sizes. See SAP Sales and Distribution benchmark and SAP BW edition for SAP HANA benchmark for complete details. For an interesting breakdown of the SAPS delivered by our Nitro-based instances, I would encourage you to explore SAP note 1656099.
  • Security: The Nitro System continuously monitors, protects, and verifies your EC2 instance’s hardware and firmware. Virtualization resources are offloaded to dedicated hardware and software, which minimizes the attack surface. Additionally, the Nitro System is locked down, prohibiting administrative access to eliminate human error or tampering.
  • Flexibility: Nitro helps us deliver all instances as truly cloud-native offerings irrespective of size. This means you can easily move from a bare metal instance to a virtual instance in a simple stop and start manner. Additionally, as you scale, there’s no need to de-provision and re-provision your storage resources, move to a new storage type, or change architecture patterns.

Summary and Getting Started

Today’s launch of additional High Memory Instances gives SAP on AWS customers more choice in how they can meet their unique requirements for cost and performance.

I highly encourage you to explore High Memory Instances and the AWS Nitro System yourself. To get started with migrating your SAP HANA workloads to an EC2 High Memory instance, see our migration guide. If you have any questions or feedback, please reach out to your account manager or SAP specialist. We remain committed to being the best place to run SAP so keep the valuable feedback coming!

Thanks,

–Steve